From January 2003 High Frequency Electronics

High Frequency Design

Copyright © 2003 Summit TechnicalMedia, LLC
BER TUTORIAL

Bit Error Rate:

Fundamental Concepts
and Measurement Issues
Gary Breed
Editorial Director

O
ne of the changes
This tutorial describes the that modern digi-
principles of bit-error rate, tal communica-
the factors that affect it, tions systems has
and the methods for its brought to radio engi-
measurement, but with a neering is the need for
recommendation that end-to-end performance
readers do further study of measurements. The mea-
its mathematical basis sure of that performance
is usually bit-error rate
(BER), which quantifies the reliability of the
entire radio system from “bits in” to “bits out,”
including the electronics, antennas and signal
path in between.
On the surface, BER is a simple concept—
its definition is simply:
BER = Errors/Total Number of Bits
With a strong signal and an unperturbed
signal path, this number so small as to be
insignificant. It becomes significant when we
wish to maintain a sufficient signal-to-noise
ratio in the presence of imperfect transmis-
sion through electronic circuitry (amplifiers,
filters, mixers, and digital/analog converters)
and the propagation medium (e.g. the radio
path or optical fiber).
Any in-depth analysis of the processes that
affect BER require significant mathematical
analysis. I will not attempt to do that in this
short introduction and overview.
Noise and BER
Noise is the main enemy of BER perfor-
mance. Noise is a random process, defined in
terms of statistics. The noise introduced by the
circuitry is described with a Gaussian proba-
bility density function, while the signal path is
usually described with a Rayleigh probability
density function. A Rayleigh, or fading, signal
path is not “noise” in the intuitive sense of the
familiar hissing sound of “white noise,” but it
is a random process that is analyzed in the
same manner as Gaussian noise.
Without going into detail, the mathemati-
cal representations of these functions allow
the analysis of a system to help predict its per-
formance. Further study of noise and statisti-
cal signal analysis is highly recommended.
Quantization errors
Quantization errors also reduce BER per-
formance, through incorrect or ambiguous
reconstruction of the digital waveform. This is
also a described by a probability function that
defines the likelihood that a digital transition
or edge detection error will occur. These errors
are primarily a function of the accuracy of the
digital-to-analog and analog-to-digital conver-
sion processes, and are related to the number
of bits used at these points in the circuit.
The accuracy of the analog modulation/
demodulation process and the effects of filter-
ing on signal and noise bandwidth also affect
quantization errors.
Eb/No and BER
BER can also be defined in terms of the
probability of error (POE),
1
POE = (1 − erf ) Eb / No
2
where erf is the error function, Eb is the ener-
gy in one bit and No is the noise power spec-
tral density (noise power in a 1 Hz band-
width). The error function is different for the

46 High Frequency Electronics

 High Frequency Design
BER TUTORIAL
each of the various modulation methods. What is more
important to note is that POE is proportional to Eb/No,
which is a form of signal-to-noise ratio.
The energy per bit, Eb, can be determined by dividing
the carrier power by the bit rate. As an energy measure,
Eb has the unit of joules. No is in power (joules per second)
per Hz (seconds), so Eb/No is a dimensionless term, or sim-
ply, a numerical ratio.
Factors Affecting BER
One way to lower the spectral noise density is to
reduce the bandwidth, but we are limited by the band-
width required to transmit the desired bit rate (Nyquist
criteria). We can also increase the energy per bit by using
higher power transmission, but interference with other
systems can limit that option. A lower bit rate increases
the energy per bit, but we lose capacity. Ultimately, opti-
mizing Eb/No is a balancing act among these factor.
48 High Frequency Electronics
BER Measurement
While the basic concept of BER measurement is sim-
ple—send a data stream through the system and compare
the output to the input—its execution is not trivial. Over
an infinitely long period of time, we can assume that a
data transmission is a random process. However, we don’t
want to wait forever to make a BER measurement! So a
pseudorandom data sequence is used for this test. We call
it “pseudo” random because we cannot create a truly ran-
dom signal using deterministic (mathematical) methods.
Fortunately, some smart mathematicians have worked
out sufficient approximations of random behavior so we
can quickly make accurate BER measurements.
The next measurement issue is a practical one—we
don’t want to completely install our system to make the
test. It would be far too costly and time-consuming to
build entire radios and install them with transmission
lines, towers and antennas, just to test the bit-error rate
performance of a particular filtering scheme or demodu-
lation circuit.
Let’s go back to the subject of noise. Radio noise is
Gaussian in its spectral power density. This includes
semiconductor junction noise (in full conduction) and
thermal noise in resistors. Specially-designed “noise
diodes” can be used to generate noise with predictable
spectral characteristics, which can be added to the signal
path in a controlled laboratory environment.
It is also possible to add Rayleigh (fading) characteris-
tics to the signal, using multiple channels with variable
time delays to simulate changing path conditions. The
combination of Gaussian noise and Rayleigh fading have
become extremely important in the development of new
digital communications equipment and systems, since
they allow convenient testing. The mathematical nature
of digital signal transmission also allows computer simu-
lation, but with the practical limits imposed by produc-
tion variations in system components.
How do we use BER data?
When BER threatens the usefulness of a system, there
a many courses of action. First, the troubleshooting pro-
cess must identify the cause of the errors. Is it circuit-
related or path related? What is the cost of the remedy?
Should we improve the hardware, pursue changes to the
transmission environment, or choose a different modula-
tion format?
Sometimes the solution is software-based error correc-
tion. These techniques have been used for many years in
terrestrial and satellite communications.These come at a
cost of lower data throughput, but they can effectively
reduce BER.
This short introduction cannot describe all issues
related to BER, but we hope it encourages you to do addi-
tional reading and study on this important topic.